iv.xb: Computes the Xie-Beni index

In lejean2000/CLAV: An R Package For Cluster Validation

Description

The Xie-Beni index is an internal cluster validation index. It defines the intercluster separation as the minimum square distance between cluster centers, and the intraclus ter compactness as the mean square distance between each data object and its cluster center. It can be used for fuzzy clustering as well. The weighted variant (default is usually the better option.

Usage

iv.xb(clustering, centers, diss, variant = "weighted")

Arguments

clustering	A vector with cluster assignments or a membership matrix.
centers	A vector with cluster center IDs.
diss	A distance matrix.
variant	Weighted (XB*) or normal (XB) formula. Check the reference for more detail.

Value

A number.

References

https://www.researchgate.net/profile/Minho_Kim7/publication/222750120_New_indices_for_cluster_validity_assessment/links/02e7e53c567697580d000000.pdf

Examples

d<-vegan::vegdist (iris[,3:4], method = "euclidean")
models = list()

for (k in 3:5) 
  models[[length(models)+1]] = fclust::FKM.med(as.matrix(d), k)

for (v in c('normal','weighted'))
  for (f in models)
    print(paste (ncol(f$U),'clusters -', v, '-', iv.xb(f$U, f$medoid, d, variant=v)))

lejean2000/CLAV documentation built on Feb. 5, 2020, 8:55 p.m.